[ExI] Meat v. Machine

[Samantha Atkins]

On Jan 1, 2011, at 9:58 AM, Eugen Leitl wrote:

> On Fri, Dec 31, 2010 at 04:19:43PM -0800, Samantha Atkins wrote:
> 
>> Until human level AGI (about 3 decades out seems to be current 
>> consensus), humans are needed.  Given that we need space based 
> 
> Do you see the difference between having a system execute a plan
> with a turnaround time of 2.5 seconds or one with half an hour?
> 
> Go buy a Kinect, put in a 2.5 second FIFO delay, and try building
> something in SL with it. Or just add a 2.5 s FIFO in the driver
> for mouse/6DOF controller, whatever use you.
> 
> Now push the FIFO to 10 seconds, 30 seconds, 1 min, 5 min, 10 min, 
> 30 min. See the difference?
> 
> Now think about what a simple collision avoidance would
> do. Just push semi-blindly, see the system settle into
> a nondesaster state. Now think *reflexes*. The dumb
> spinal cord is on the Moon, 2.5 sec turnaround time away,
> your brain is here.
> 
> Everyone seems to think it's realtime fine-motorics,
> or bust. Not so. Yes, there's a difference between 30 ms
> and 2500 ms. But there's a much larger difference between
> 2500 ms and 1800000 ms. That's one hell of a handicap,
> even without microgravity.

OK.  Design the robotics that can, say, repair the Hubble, do various space walk equivalent missions all with no humans closer than earth.   Oh, the systems must be general enough to be used for anything a trained human can do as far as physical capabilities are concerned.   All but rudimentary control you mention above is done remotely.  Then get back to me.   

> 
>> resources before three decades from now we must build out human 
>> support local space/lunar infrastructure.    
> 
> Humans are irrelevant. At least when it comes to space.
> You want to go places, you have to stop wearing the
> stupid man suit.

I just presented an argument why they are not yet irrelevant than you have not countered successfully.    The only general intelligence of sufficient power currently around does not yet have the ability to shed its biology.  So again, if you need more localized general intelligence rather than at tens or thousands of miles remove then you need humans in space - today.

> 
>> You need a lot of high mass initial equipment to lift from 
> 
> I disagree that you need to launch large (100 ton)
> packages. I think you can work well with >100 kg
> packages. With plasma thrusters you can probably
> deliver one half to one third of LEO payload to 
> Moon surface semi-softly. So a ton to LEO is a 
> useful threshold.

Construction materials?  Large focusing antennae for SSP projects?  You can either do hundreds or thousands of launches or you can do a relatively few large launches for the acceleration hardened larger components.  The latter is cheaper in all ways and gets a larger resource base in play much more quickly. 

> 
>> the gravity well in any any case to have a basis to built 
>> from this side of mature nano-assembler seeds which are 
>> at least 5 - 6 decades out.     It is a good question what 
>> the minimal amount of lift needed is given the current tech 
> 
> We're well in excess of what we need. It would be nice
> if prices would come down a bit, but that is not actually
> relevant.

I don't see why you would claim that.  Many projects are not doable given today's launch cost and launch facility limitations.  

> 
> More importantly, you can start working now, as none of the
> parts rely on particular features of transport system you're
> going to use 15-20 years from now.
> 

Which parts for precisely what?


>> state of the art over time.  The amount of mass you need 
>> to lift from earth in inversely proportional to the 
>> sophistication of the technology.     But it is today quite substantial.   
> 
> I disagree it is substantial. And the only way to know is
> to start working *now*, so that in 15-20 years you have all 
> the numbers.
> 


On what?  What do you suggest launching that is off the shelf now and for what purposes?

- s